In the previous chapter, we learned some basic concepts to draw line and scatter plots with Matplotlib, and made adjustments to a few elements. Now that we are familiar with the Matplotlib syntax, we are ready to go further and explore the potential of Matplotlib.
In this chapter, we will discuss:
Color is an essential element in any visual. It can have a huge impact on how graphics are perceived. For example, sharp color contrast can be used to highlight a focus; a combination of several distinct colors is useful in setting up a hierarchy.
In Matplotlib 2, colors have been set by default to better differentiate between categories; and to perceive continuous numerical values more intuitively, yet we often need better control over the colors to represent our data. In this section, we will introduce the common color options in Matplotlib.
A color cycle is a list of colors used to control the color of a series of elements automatically, such as each data series in multiline plots. In Matplotlib 2.0, the default color cycle has expanded from 7 to 10 colors using the category10 palette in Data-Driven Documents (D3) https://github.com/d3 and Vega, a declarative language for visualization grammar. These colors are designed to show good contrast between...
We have demonstrated how to draw line plots and scatter plots in the previous chapter. We know that scatter plots are made up of dots denoting each data point, whereas line plots are generated by joining dots of data points. In Matplotlib, the marker to mark the location of data points can be customized to have different styles, including shape, size, color, and transparency. Similarly, the line segments joining the data points as well as different 2D lines that share the same class in the object-oriented Matplotlib structure can have their styles adjusted, as briefly demonstrated in the grid section of the previous chapter. Adjusting marker and line styles is useful in making the data series more distinguishable, and sometimes for aesthetic considerations. In this section, we will go through the details and implementation methods of marker and line styles in Matplotlib.
Besides the most basic scatter and line plots, Matplotlib provides a versatile collection of plot types to serve different data visualization purposes. In this section, we will introduce the rationale of plot type selection and the usage of each type.
A successful visualization must communicate the message well. In order to achieve this goal, we need to have a good understanding of the nature of our data as well as the advantages and limitations of each plot type in illustrating different relationships in data. In choosing the right plot type to display, we have the following considerations:
To enhance the understanding of plot details, we may sometimes add in text annotations for explanation. We will now introduce the methods of adding and adjusting text in Matplotlib plots.
We can add text to our plot by calling plt.text(x,y,text); we specify the x and y coordinates and the text string.
Here is a quick example:
plt.text(0.25,0.5,'Hello World!',fontsize=30) plt.show()
You can see in this figure the Hello World! message appearing in the center of the plot:
We have learned to style our plots step by step so far. For more persistent and portable settings, we can apply a predefined global style via the matplotlib.style module:
## Available styles Matplotlib provides a number of pre-built style sheets. You can check them out by with `matplotlib.style.available`. import matplotlib as mpl mpl.style.available Out[1]: ['seaborn-talk', 'seaborn-poster', '_classic_test', 'seaborn-ticks', 'seaborn-paper', 'ggplot', 'seaborn', 'seaborn-dark', 'seaborn-bright', 'seaborn-pastel', 'fivethirtyeight', 'Solarize_Light2', 'classic', 'grayscale', 'bmh', 'seaborn-dark-palette', 'seaborn-whitegrid', 'seaborn-white', 'dark_background', 'seaborn-muted', 'fast', 'seaborn-notebook', 'seaborn-darkgrid', 'seaborn-colorblind', 'seaborn-deep']
As visualization is about delivering messages, the more we think from the reader's perspective, the more effective it will be. An attractive graphic catches more attention. The easier to read a plot is, the more likely are readers to understand the message. Here are several basic principles in designing data plots.
The hierarchy can use no more than three levels of font family, weight, and sizes. Use less fancy font families, Sans Serif font if possible. Make sure the font size is large enough to be legible
Serif versus Sans Serif Serif means decorative edges on alphabets. And sans means without in French. As the names imply, Sans Serif fonts are plainer and more simplistic than Serif fonts in general. Let's take the most popular examples of default fonts in Microsoft Office. Times New Roman used in Office 2007 and before is a Serif font, whereas the newer Calibri is a Sans Serif font.
Congratulations! You have now mastered the most commonly used plots and the basic methods to customize plots. We are now ready to move on to more advanced Matplotlib usage.
In the next chapter, we will cover more plot types with the help of third-party packages, methods to optimize displays for multiple plots and axes in certain scales, as well as showing pixels in images. Stay tuned!
© 2018 Packt Publishing Limited All Rights Reserved
